Re: ib - r31623 - abiword/trunk/src/af/gr/xp

Ingo,

Obviously my patch cannot solve a font mapping issue (i.e. using a
Symbol font with an improper
conversion to Unicode). As I stated in my first e-mail on the subject,
the solution is to

1) stop using the dingbats and symbol fonts to represent bullets
2) use standard unicode code points
3) let pango handle font substitution if necessary.

As you mentioned, the 2nd point is effectively implemented in the code
(at this point we are setting the characters for the bullets in
fl_AutoNum.cpp at non-unicode values and resetting them to unicode
values in gr_CairoGraphics.cpp). My patch fixes the 3rd point (and
fontconfig does font substitution also). But point 1 still needs to be
addressed. Basically, our source code should not explicitly use any
non-unicode font (only keeping font remapping functions such as
adobeToUnicode and adobeDingbatsToUnicode).

Simon

On Fri, Aug 3, 2012 at 6:16 AM, Ingo Brückl <ib@wupperonline.de> wrote:
>
> Simon Larochelle wrote on Thu, 2 Aug 2012 08:45:47 -0400:
>
>> I made a patch to do font substitution on fields. I attached the patch
>> to Bug 12334 if someone wants to check it. I will commit it to the SVN
>> tomorrow if there are no problems.
>
> Here is my report. I've checked in a "clean" environment, i.e. reversion of
> all Standard Symbols L stuff from AbiWord source, no fontconfig mapping for
> Dingbats and removal of font Standard Symbols L. I still have the TTF Symbol
> which I need for different purposes.
>
> Short version: It doesn't improve things much. For "Dingbats", the same can
> be achieved with a fontconfig font mapping, but the patch seems preferable.
>
> Long version: 11 of 12 bullets are ok, but the first one (the bullet
> character) is still the infinity character (which happens to be the glyph in
> TTF Symbol there). An other problem is the code in gr_CairoGraphics.cpp after
> line 2500, handling "symbol". This even affects xap_UnixDlg_Insert_Symbol
> where the glyphs of TTF Symbol didn't show up correctly. (When trying to fix
> this by the patch of r31623, the implies bullet glyph gets wrong as well.)
>
> The main problem remains. We are demanding font Symbol for two of the
> bullets, although the font we are referring to is really called Standard
> Symbols L. (It seems that was a fontmap mapping once from Standard Symbols L
> to Symbol though.)
>
> Although everyone is objecting, when there is font Symbol mentioned or used
> for bullet lists in AbiWord it rather should be Standard Symbols L, because
> that is the name of the font that is meant (and that seems the reason for the
> code introduced in r18148 and later commented in r27532).
>
> Ingo
Received on Fri Aug 3 18:01:04 2012